The COVID-19 pandemic has been accompanied by wrenching images, from previously healthy people hooked up to ventilators to photos of normally packed plazas and boulevards standing almost empty. But in the United States, perhaps none have been so widely shared – and derided -- as pictures of college students partying on then still-open beaches in Florida, celebrating spring break. As one partier infamously put it on March 18: “If I get corona, I get corona. At the end of the day, I’m not gonna let it stop me from partying.”
All of this led to outrage on social media, along with loud demands that Florida Gov. Ron DeSantis close the beaches and put spring break effectively out of business. The governor refused. This has led to confident predictions that Florida had been seeded with coronavirus cases and was likely to emerge as the next hot spot.
There’s more to be said about the dangers of overconfident predictions – among both professionals and pundits. We should also remember that the story of this pandemic is hardly written yet; we’re still in the opening chapters. But for now, the data don’t seem to support a spring break-fueled explosion in the Sunshine State. Even if it is too early to give a definitive answer, the answer to the question right now isn’t as clear as I had expected it would be three weeks ago.
So what do we actually see in the data (which are admittedly imperfect, about which more below)? If spring break was really creating an epidemic in Florida, we should expect to see a “bending of the curve” – in the wrong direction – starting to emerge about five days after spring break. This curve-bending should also be more severe than what we see occurring in most other states, assuming that what Florida did was uniquely dangerous.
I’ve taken the data from the indispensable Johns Hopkins University Center for Systems Science and Engineering dashboard, which tracks outbreaks of the virus by nation, state and county. The data are current as of April 7, 2020. I’ve then traced the number of reported cases by state dating back to March 15, a few days before the spring break stories broke. This is useful because cases that were seeded by spring break partiers largely should have presented by now as well as many of the second-order transmissions.
As you can see, New York and, to a lesser extent, New Jersey dominate the caseload in the United States. You may not be able to read the state names in the bottom right (click on the chart to enlarge it), but Florida is in the cluster well beneath New York and New Jersey. Whatever else we might say about Florida, it isn’t the next New York, at least not yet.
We can see this better if we eliminate New York and New Jersey from the chart. We’ll also place the states outside of the cluster with Florida in the background.
As you can see, Florida isn’t doing anything unusual here; it’s tracking other states in its cluster, including ones that aren’t being heavily criticized for their policies, such as Pennsylvania, Illinois and California. Things get a bit grainy on the left side of the chart – maybe Florida was actually well toward the bottom of state infections and then surged? No, not so much:
This chart traces the ranking over time of the states. The states with the 10 most cases as of April 7 are highlighted. This allows us to see whether a state has surged relative to other states. As you can see, the data are pretty noisy as of mid-March, when small changes in the number of cases a state had could have a substantial impact on whether it had the 30th most cases or 40th most.
Regardless, Florida’s relative position has been remarkably stable during this time period. It has fluctuated between having the fifth most cases (multiple times) and the ninth most cases (March 16, 25, and 28). Regardless, if you want to know what a spring break surge might look like, check out Louisiana (which goes from 41st in cases to fifth over this time period, possibly due to Mardi Gras) and Michigan (which goes from 43rd to third, for who knows what reason). So far, at least, the trend in Florida isn’t consistent with an exploding caseload.
Even this might not be fair to Florida. It is, after all, the third most populous state in the nation. We’d expect it to have more cases than, say, Wyoming, even if it were doing a good job containing the virus. We can redo the above charts examining cases per million residents:
On a per capita basis, we see a bit more separation. New York, New Jersey, Louisiana, Massachusetts, Connecticut and Michigan stand out as states with caseloads increasing at unusual rates. Florida, however, is buried in the middle of the pack. We can see this better by removing the preceding six states.
Florida is wedged between Nevada and Mississippi, along with Idaho. Again, we’re not seeing the sort of exponential growth that we would expect to see had the media meltdown over spring break been fully justified. Finally, after fluctuating wildly early on (at a time where a fraction of a case per million could move a state more than 10 ranks), Florida has stabilized in the middle of the pack in terms of rank.
Overall, Florida (so far) doesn’t seem to be showing the sort of unusual acceleration that we would expect if it were really en route to becoming the next New York.
Having said that, this article is headlined with a question -- for a couple of reasons. First, this pandemic is not over. Florida could explode tomorrow. It’s probably too early to answer the question, but we should at least be willing to contemplate it. Second, there are reasons to push back against the above analysis. We turn to those reasons now, but they illustrate a couple of broader points about the way we should think, and write, about the pandemic.
For one thing, some critics suggest that these numbers mostly reflect a low level of testing in Florida. To put things in perspective, when South Korea was the envy of the world for its testing prowess, it was testing 10,000 people per day and had tested a total of around 250,000 people. That would translate to testing about 64,000 people per day in the United States (which has about 6½ times the population of South Korea) and a total of about 1.6 million Americans. As of today, according to the COVID tracking project, the United States has performed over 2 million tests and is testing around 137,000 people per day. Even this doesn’t tell the entire story as some states have had almost 80% of their tests performed in private labs.
One can argue about these statistics, just as one can assert that the United States should still be doing better. But if your image of America is of a country flying blind with no testing, as was more or less the case for much of March, you’re in need of an update. Regardless, Florida’s data probably aren’t corrupted by a dearth of testing, as the differences among states in testing rates aren’t that great. The overall utility of this approach may be questionable, as states with higher levels of sickness would request more tests, but it’s the best we have. Here’s the percentage of the population tested in each state.
Half of all states have tested between 0.45% and 0.86% of the population – a very narrow band. The overall range is just 1.5%. Florida’s hasn’t tested the most people in the country, but it hasn’t tested the fewest either. It’s tested a larger share than states generally recognized as doing a good job with the virus, such as Ohio and California. It’s tested more than states with large outbreaks, such as Michigan. Overall, it seems hard to explain away the fact that New York currently has 11 times as many cases per capita as Florida on the grounds that New York has tested three times as many people per capita as Florida.
Nonetheless, a friend I know and trust (and who has taught me a lot about this outbreak) suggests that beachgoing college students have gone back home after spring break ended and have seeded their states with this nasty virus. This probably happened, at least to some extent, and my point isn’t that there were zero costs associated with letting spring break move forward. But it seems unlikely that the people who infected vacationers in Florida wouldn’t have also infected native Floridians and other locals on the beach, including bartenders, hotel staff, and so forth. That, after all, is the storyline about how spring break was supposed to create an explosion in Florida. Some of this probably occurred, but for whatever reason, it hasn’t yet led to a Florida explosion.
You can look at the locations where the virus is picking up in Florida and infer that the counties are also spring break hotspots -- though you would presumably have to explain the large number of cases in Sumter County (The Villages) via a different mechanism -- but these are also the more urbanized areas that typify viral spread in other states. We’d expect Miami-Dade to have an outsized number of cases even without spring break. These are also, incidentally, places where New Yorkers fleeing the city were landing, which makes it all the more surprising that Florida isn’t following New York’s trajectory (yet). Regardless, the variation in cases among counties in Florida isn’t that large; no county yet has more than 10 cases per 10,000 residents, with most of the counties with at least one case clustered between two and five cases per 10,000 residents. Incidentally, Volusia County, which includes Daytona Beach, is on the lower end of that range even though it didn’t close its beaches until Friday.
There were also reasons from the start to be skeptical of the spring break explosion storyline: for example, that DeSantis had already banned gatherings in excess of 10 people. Most of the infamous pictures of crowded beaches were taken from far away, but if you look closely most of the clusters of people aren’t really that close. Perhaps most importantly, there’s evidence that heat slows the spread of the virus. If you had to have people not socially distancing, a hot, sunny beach without smooth surfaces and with lots of salt water is probably a good place to have it happen. Of course, part of the problem with spring break isn’t what happens on the beaches – indeed the worst social distancing violations probably happen off the beaches. But the point isn’t that there were zero transmissions during spring break. It’s simply that we don’t see the out-of-control transmission some expected.
Which brings me to my final point. In response to this information, a person might argue that even if Florida didn’t get a New York-style explosion in cases, it probably has more cases than if it had closed down the beaches and theme parks earlier. I’m inclined to agree. But I’m not sure that avoiding what seems to be at best a fairly marginal (for now) increase in cases beyond what the state would have suffered is clearly worth the damage it would have done to local businesses by cutting off spring break. In other words, DeSantis might have gotten the cost-benefit right.
To be clear, I support most of the social distancing that has been imposed (I think the jury is still out on school closures). I don’t think the cure is worse the disease right now. Analogies to the flu are too flippant; left unchecked this virus seems certain to kill off at least one order of magnitude more people than seasonal flu. At the same time, the social distancing seems to have stopped a pretty nasty flu outbreak dead in its tracks. We could do this every winter and save tens of thousands of lives annually; over the course of my lifetime the failure to do this every winter probably will kill more Americans than unchecked COVID-19 spread would. We don’t do that, of course, but to be clear: We don’t do it because the social and economic costs of doing so would be too great.
There’s a certain beauty and moral rightness in saying that every life counts, and that one life lost to this virus is one life too many. In reality, almost every one of us at some point solves the cold equations against life. There is more than ample room to conclude that DeSantis did it wrong here, but we should be honest with ourselves that we all have our limits as to what we will tolerate, and be willing to consider arguments to the contrary if things don’t change for the worse there.